Is Secured Debt Used to Redistribute Value From Tort Claimants in Bankruptcy - An Empirical Analysis

8/13/2019 Is Secured Debt Used to Redistribute Value From Tort Claimants in Bankruptcy - An Empirical Analysis

1/45

Yale Law School

Yale Law School Legal Scholarship Repository

Faculty Scholarship Series Yale Law School Faculty Scholarship

1-1-2008

Is Secured Debt Used to Redistribute Value fromTort Claimants in Bankruptcy? An Empirical

AnalysisYair ListokinYale Law School

is Article is brought to you for free and open access by the Yale Law School Faculty Scholarship at Yale Law School Legal Scholarship Repository. It

has been accepted for inclusion in Faculty Scholarship Series by an authorized administrator of Yale Law School Legal Scholarship Repositor y. For

more information, please [email protected].

Recommended CitationListokin, Yair, "Is Secured Debt Used to Redistribute Value from Tort Claimants in Bankruptcy? An Empirical Analysis" (2008).Faculty Scholarship Series. Paper 554.hp://digitalcommons.law.yale.edu/fss_papers/554
http://digitalcommons.law.yale.edu/http://digitalcommons.law.yale.edu/fss_papershttp://digitalcommons.law.yale.edu/fssmailto:[email protected]:[email protected]://digitalcommons.law.yale.edu/fsshttp://digitalcommons.law.yale.edu/fss_papershttp://digitalcommons.law.yale.edu/


2/45

S SECURED DEBT USED TO REDISTRIBUTEVALUE FROM TORT CLAIMANTS IN

BANKRUPTCY AN EMPIRICAL ANALYSISYAIR LISTOKINt

ABSTRACTMany scholars question the priority enjoyed by secured debt in

bankruptcy. They fear that secured debt will be used to inefficientlyredistribute value away from preexisting unprotected creditors offirm. These scholars advocate a host of legal innovations, such as superpriority or tort claimants with respect to other creditors, tomitigate the redistributionalproblem. Other scholars minimize theredistributional roblem, however, and argue thatpriority or securedcredit is efficient. To help resolve this debate, this Article examines theredistributional heory from an empiricalperspective. In particular, tfocuses on secured debt usage by publicly tradedfirms facing largetort liabilities ( high-tort firms). In theory, secured debt should beattractive for high-tort firms because they have a large class ounsecured and uncovenanted creditors (tort claimants) exposed toredistribution in bankruptcy through the use of secured credit. TheArticle's empirical analysis contradicts the redistributional heory'sprediction, however. High-tortfirms have unusually low amounts ofsecured debt. Although this result is very difficult to explain under theredistributional heory, it can readily be explained according to othertheories of secured debt. Several important policy implications forbankruptcy prioritiesollow from these findings.

Copyright 2 8 by Yair Listokin. Associate Professor of Law Yale Law School. I would like to thank Barry AdlerKenneth Ayotte Ian Ayres George Priest Alvin Klevorick Roberta Romano Alan Schwartz

Kathryn Spier George Triantis and participants at the Conference on Empirical Legal Studiesat the University of Texas School of Law for many helpful comments and suggestions. All errorsare my own


3/45

DUKE LAW JOURNALTABLE OF CONTENTS

Introduction 1038I. Explaining the Use of Secured Debt and EvaluatingSecured Debt's Efficiency 1046

A. The Costs of Secured Debt 1047B. The Benefits of Secured Debt 10481. Controlling Monitoring and Agency Costs 10482. Redistribution-Priority Related 1051C. The Normative Implications of Theories of

Secured Debt for Bankruptcy Priorities 1053II. Empirical Analysis: Secured Debt and Tort Liability 1055

A. Theories of Secured Debt and Predictions RegardingSecured Debt Usage and Tort Liability 1056

B. Identifying Firms with Large Tort Liabilities 1057C. Financial Data and Summary Statistics 1059

III. Statistical Analysis of the Relationship betweenTort Risk and Secured Debt Usage 1062

A. Simple Statistical Analysis of Secured Debt Usage 1062B. Regression Analysis of Secured Debt Usage:

Tobit Model 1065C. Tim e-Series A nalysis 1071

1. Tobacco Industry Fixed Effects ...................... 10722. Secured Debt Usage and the Approach of Bankruptcy 1073IV. Evaluation of the Statistical Results 1076C onclusion 1078

INTRODUCTIONSecured creditors enjoy priority status in bankruptcy with respect

to other creditors.' Although the rule is well established, itsdesirability is the subject of decades of scholarly debate.2 Some

1 See 11 U.S.C. 361-364 (2000 Supp. V 2005). The priority of secured creditors isprotected by the adequate protection clauses of these sections, which ensure that holders ofcollateral whose repossession has been stayed by bankruptcy should receive the full amount oftheir secured claim by the end of the bankruptcy process.

2. Professor Robert Scott's invaluable law review article dates the debate to 1979. RobertE. Scott, The Truth About Secured Financing 82 CORNELL L. REV. 1436, 1437 n.1 1997)[hereinafter Scott, Truth About Secured Financing]. For other articles developing the debate,see generally Barry E. Adler, An Equity-Agency Solution to the Bankruptcy-PriorityPuzzle 22J LEGAL STUD. 73 1993); Douglas G. Baird, The Importanceof Priority 82 CORNELL L. REV.1420 1997); Lucian Arye Bebchuk Jesse M. Fried, The Uneasy Case for the Priority ofSecured Claims in Bankruptcy 105 YALE L.J. 857 1996) [hereinafter Bebchuk Fried, Uneasy

[Vol. 57:1037038


4/45

2008] BANKRUPTCY 1039academics assert that priority for secured debt promotes inefficientuses of secured debt as a means of redistributing value away fromunsecured creditors toward those with collateral. Late-arrivingsecured creditors can leapfrog earlier unsecured creditors,redistributing value to the benefit of the issuer and the securedcreditor but to the detriment of unsecured creditors and, possibly, tosocial welfare (in the Kaldor-Hicks sense). Other scholars minimizethe salience of the redistributive motive and claim that the priority ofsecured debt mitigates agency conflicts between borrowers and

Case 1]; Lucian Arye Bebchuk Jesse M. Fried, The Uneasy Case for the Priority of SecuredClaims in Bankruptcy: FurtherThoughts and a Reply to Critics,82 CORNELL L. REV. 1279 (1997)[hereinafter Bebchuk Fried, Uneasy Case 2]; F.H. Buckley, The Bankruptcy Priority Puzzle,72 VA. L. REV 1393 (1986); David Gray Carlson, Secured Lending as a Zero-Sum Game, 19CARDOZO L. REV. 1635 (1998); Steven L. Harris Charles W . Mooney, Jr., Measuring theSocial Costs and Benefits and Identifying the Victims of Subordinating Security Interests inBankruptcy, 82 CORNELL L. REV. 1349 (1997) [hereinafter Harris Mooney, Measuring theSocial Costs]; Steven L. Harris Charles W . Mooney, Jr., A Property-BasedTheory of SecurityInterests: Taking Debtors Choices Seriously, 80 VA. L. REV. 2021 (1994) [hereinafter HarrisMooney, Property-Based Theory]; John Hudson, The Case Against Secured Lending, 15 INT LREV. L. ECON. 47 (1995); Thomas H. Jackson Anthony T. Kronman, Secured FinancingandPriorities Among Creditors, 88 YALE L.J. 1143 (1979); Hideki Kanda Saul Levmore,Explaining Creditor Priorities,80 VA. L. REV. 2103 (1994); Saul Levmore, Monitors andFreeriders n Commercialand CorporateSettings, 92 YALE L.J. 49 (1982); Lynn M. LoPucki, TheUnsecured Creditor sBargain, 80 VA. L. REV. 1887 (1994); Ronald J. Mann, Explaining thePatterno Secured Credit, 110 HARV. L. REV. 625 (1997); Randal C. Picker, Security Interests,Misbehavior,and Common Pools,59 U. CHI L. REV. 645 (1992); Steven L. Schwarcz, The EasyCase for the Priorityof Secured Claims in Bankruptcy, 47 DUKE L.J. 425 (1997); Alan Schwartz,The Continuing Puzzle of Secured Debt, 37 VAND L. REV. 1051 (1984) [hereinafter Schwartz,The Continuing Puzzle]; Alan Schwartz, Priority Contracts and Priority in Bankruptcy, 82CORNELL L. REV. 1396 (1997) [hereinafter Schwartz, Priority Contracts]; Alan Schwartz,Security Interestsand Bankruptcy Priorities:A Review of Current Theories, 10 J. LEGAL STUD(1981) [hereinafter Schwartz, Bankruptcy Priorities];Alan Schwartz, Taking the Analysis ofSecurity Seriously, 80 VA. L. REV. 2073, 2075-81 (1994) [hereinafter Schwartz, Taking theAnalysis of Security Seriously]; Robert E. Scott, A Relational Theory of Secured Financing,86COLUM. L. REV. 901 (1986) [hereinafter Scott, RelationalTheory]; Paul M. Shupack, Solving thePuzzle of Secured Transactions,41 RUTGERS L. REV. 1067 (1989); George G. Triantis, A Free-Cash-Flow Theory of Secured Debt and Creditor Priorities, 80 VA. L. REV. 2155 (1994)[hereinafter Triantis, A Free-Cash-Flow Theory]; George G. Triantis, Secured Debt UnderConditionsof Imperfect Information,21 J. LEGAL STUD. 225 (1992); Elizabeth Warren, MakingPolicy with Imperfect Information: The Article 9 Full Priority Debates, 82 CORNELL L. REV.1373 (1997); James J. White, Efficiency Justifications or Personal Property Security, 37 VAND.L. REV. 473 (1984).

3. See, e.g. Bebchuk Fried, Uneasy Case 1 supra note 2, at 859; Bebchuk Fried,Uneasy Case 2 supra note 2, at 1314-15; Hudson, supra note 2, at 47; LoPucki, supra note 2, at1891; Warren, supranote 2, at 1389-90.


5/45

DUKE L W JOURN L

lenders and facilitates efficient loans that could not occur if securedcredit were not awarded priority.4

In a law review article, Professor Robert Scott summarized thedebate as follows:To some extent, [the] leverage [afforded by secured debt] seems tobe a singularly useful means of reducing conflicts of interest inherentin financial contracting relationships. These benefits are efficiencyenhancing. To some degree, however, [the] leverage [afforded bysecured debt] also appears to be a singularly useful means ofenhancing the creditor's probability of repayment relative to othercreditors. If as seems plausible, some or many) of these othercreditors do not adjust to this reduction in bankruptcy share, there isa redistributional benefit to the creditor that the debtor does notfully internalize in assessing its total interest bill. This, then, wouldlead to some inefficient uses of security as well as raise problems ofdistributional fairness). The question, in short, is simple: What arethe relative values of these two offsetting effects? At this point wedo not have a clue.5This Article attempts to get a clue by testing the predictive

value of the redistributional theory of secured debt. If redistributionconstitutes a principal motive for firms' use of secured debt, thenfirms with greater opportunities for redistribution should issue moresecured debt than other firms.6 In particular, firms facing outsized,noninsurable tort liabilities should issue large amounts of secureddebt. Tort claimants are the paradigmatic nonadjusting unsecuredcreditor.7 Tort claimants cannot demand covenants to preventleapfrogging by later secured creditors, nor can they demand aninterest rate premium in lieu of covenant protection.8 Unlike other

4. See e.g. Carlson, supra note 2, at 1643-46; Harris Mooney, Measuring the SocialCosts supra note 2, at 1350; Harris Mooney, Property-BasedTheory supra note 2, at 2021-22;Kanda Levmore, supra note 2, at 2111-14; Schwarcz, supra note 2, at 432-33; Schwartz,Priority Contracts supra note 2, at 1397 98

5. Scott, Truth About Secured Financing upra note 2, at 1461.6 See Schwartz, Bankruptcy Priorities upra note 2, at 30 (stating that the distributional

explanation predicts that firms will issue secured debt only when a substantial number of theircreditors are uninformed ).

7. For example, tort claimants are listed first in the taxonomy of nonadjusting creditorsdescribed by Bebchuk and Fried. Bebchuk Fried, Uneasy Case1 supranote 2, at 882.

8. Late-arriving tort creditors-those who bring their successful suits in or just nearbankruptcy-can obviously not obtain before their claim is brought to judgment or settled.Known tort creditors may sell their claims to other parties. The price at which these claims willbe sold will be discounted to reflect the ability of the company to adulterate the value of the tort

1040 [Vol. 57:1037


6/45

BANKRUPTCYcreditors, they are exogenously unsecured (by virtue of the legalsystem). Firms facing large (and potentially terminal) liabilitiesshould capitalize on this vulnerability by issuing large amounts ofredistributive secured debt. The tortfeasing firm would pay lowerinterest rates on secured debt than on unsecured debt because asecured creditor will be paid in full before tort claimants receive anydistribution should the tort liability ultimately force the company intobankruptcy.

Other rationales for secured debt make sharply contrastingpredictions for the amount of secured debt likely to be issued by firmsfacing large tort liabilities ( high-tort firms ). Although such firmsare at risk of bankruptcy, they are unlikely to experience the agencyproblems thought to characterize secured debt issuers under alternatetheories of secured debt. For example, secured debt is unlikely toprevent a high-tort firm from risk shifting toward riskier projectsbecause the firm's relatively stable income is not the source of thebankruptcy risk; instead it is the firm's tort liabilities that are theproblem. Similarly, firms facing large tort liabilities, such as tobaccofirms, are not the type of firms likely to need secured debt as acommitment device.9 Large tobacco firms can credibly commit toavoid debt dilution through the use of covenants, making costlysecured debt unattractive as a commitment device. Finally, high-tortfirms may produce stable cash flows-limiting the attractiveness ofthe financial slack proffered by secured debt.'Consequently, high-tort firms offer an ideal test of the predictivevalue of the redistributive theory of secured debt. If redistribution is aprime motive for secured debt, then such firms should issue largeamounts of secured debt. If redistribution is relatively unimportant,then firms facing large tort liabilities should not issue unusualamounts of secured debt.The answer to the question of whether or not high-tort firmsissue large amounts of secured debt has normative implications formany of the debates raging around the priority of secured debt. First,many commentators assuming that redistribution from tort claimants

claim by issuing secured debt. Thus, the value of any tort claim that has not been paid in full isaffected by the possibility of redistribution.

9 See Schwartz, Priority Contracts supra note 2, at 1412 suggesting that borrowers whocannot credibly commit to obey nondilution covenants may choose to issue secured debt).

10. See George G. Triantis, FinancialSlack Policy and the Laws o Secured Transactions.29 J. LEGAL STUD 35, 35-37 2000 .

2008] 1041


7/45

DUKE LAW JOURNALoccurs, have argued for a change in legal regime toward one awardingsuperpriority to tort claimants with respect to other creditors. If tortliabilities lead to redistributive secured debt issuance, then priorityfor secured creditors with respect to tort claimants is both unfair andinefficient. Secured creditor priority is unfair because tort claimantsreceive less than they are entitled to and cannot receive a higheraward to compensate them for the risk that secured claimants willsubordinate their claims in bankruptcy. It is inefficient because itdilutes the incentives for potential tortfeasing companies to takeprecautions against injuries and encourages such companies to usesecured debt in situations in which, redistribution aside, they wouldprefer to issue unsecured debt. As a result, superpriority for tortclaimants makes good sense-assuming that firms actually usesecured debt to expropriate value from tort claimants. 2

Similarly, other commentators have pointed to redistribution as arationale for adjusting the priority of secured credit more generally.Professors Lucian Arye Bebchuk and Jesse M Fried explain that tortclaimants are not the only nonadjusting creditors. Other nonadjustingcreditors include trade creditors and all prior unsecured creditors of adebtor who is considering borrowing on a secured basis.'3 Becausesecured credit redistributes value away from these third parties,secured debt may exceed the socially efficient amount. 4 To prevent

11. E.g. Barry E. Adler, World Without Debt, 72 WASH. U. L.Q. 811, 826 1994);Kathryn R. Heidt, CleaningUp Your Act: Efficiency Considerationsn the Battle for the Debtor sAssets in Toxic Waste Bankruptcies,40 RUTGERS L. REV. 819, 851-63 1988); Rebecca J. Huss,Revamping Veil Piercing or All Limited Liability Entities:Forcing the Common Law Doctrineinto the StatutoryAge, 70 U. CIN. L. REV. 95, 133 2001) (arguing that tort claims should be givensuperpriority in bankruptcy because tort creditors have no ability to allocate risks or demandsecurity); David W. Leebron, Limited Liability, Tort Victims, and Creditors,91 COLUM L. REV.1565, 1643 1991); LoPucki, supra note 2, at 1913; Note, Switching Priorities:Elevatingthe Statusof Tort Claims in Bankruptcy in Pursuito Optimal Deterrence,116 HARV. L. REV. 2541, 25622003). See generally Andrew Price, Note, Tort Creditor Superpriorityand Other ProposedSolutions to Corporate Limited Liability and the Problem of Externalities, 2 GEO. MASON L.REV. 439 1995).

12. Awarding superpriority to tort claimants may prove to be sound policy even if secureddebt is not used to redistribute value away from secured creditors. The arguments presentedhere and elsewhere in favor of superpriority will not apply, however. Instead, scholars mustformulate new arguments in favor of the change.

13. These nonadjusting creditors are different from tort creditors in that they may demanda contractual premium in exchange for accepting unsecured status. Once they sign theircontracts, however, they instantly become nonadjusting.

14. This argument alone explains too much, as it implies that all credit should be secured.See Adler, supra note 2, at 74; Schwartz, Bankruptcy Priorities, upra note 2, at 30-33. As a

1042 [Vol 57:1 37


8/45

BANKRUPTCYinefficient use of secured debt, Bebchuk and Fried, among others,advocate deviations from full priority for secured credit in favor ofadjustable rules that enable sharing of assets between secured andunsecured creditors. 5 Again, however, if redistribution is not anempirically important phenomenon, then the impetus for this changedisappears.

Finally, a finding that high-liability companies issue secured debtwould provide evidence to support the claim that companies altertheir capital structure to avoid tort liability.6 If companies behave thisway, then fears about the death of liability are more than idleworries and reforms to the tort system will be essential to preventcompanies from avoiding liability through legal manipulations.7 Ifcompanies do not use secured credit to avoid liability, however, thenit is unlikely that companies will favor more radical rearrangementsto capital structure when the benefits of avoiding liability are morenebulous than the case of secured credit. In total, a finding thatcompanies facing large tort liabilities load up on secured debtwould have important implications for the future direction of the lawat the intersection of bankruptcy, corporate law, torts, and securedtransactions.

Although the stakes are high, there has been little empiricalinvestigation of the use of secured credit to redistribute value awayfrom tort creditors. Indeed, one scholar decried the general lack [of]any persuasive empirical data in the secured debt literature.'Although several financial and legal scholars have investigated theuse of secured lending, these papers have either ignored theredistributive motive or focused on the use of secured credit for

result, Bebchuk and Fried include redistribution as one of several explanations for secured debt.Bebchuk Fried, Uneasy ase 1 supra note 2, at 880-904.

15. Even if these nonadjusting creditors receive a premium for the risk of redistribution,secured debt will be used too frequently from an efficiency standpoint, with a resultantdeadweight loss. See Bebchuk Fried, Uneasy ase 1 supra note 2, at 880-904.

16. See generally Richard R.W. Brooks, iability and OrganizationalChoice 45 J.L.ECON. 91 2002) (finding that companies expanded their boundaries in response to the threat ofhigher liability from environmental damages); Al H. Ringleb Steven N. Wiggins, Liability andLarge-Scale Long-Term Hazards 98 J. POL. ECON. 574 1990) (finding that exposure tosignificant liability leads to smaller, judgment-proof companies).

17. See generally Lynn M. LoPucki, The Death of Liability 106 YALE L.J. 1, 4 1996)(noting that if judgments of liability cannot be enforced, then liability is merely symbolic ).

18. Scott, Truth About Secured Financing supra note 2, at 1437; s also Warren, supranote 2, at 1374 (noting that empirical questions have not been addressed in any detail).

2 8] 1043


9/45

DUKE LAW JOURNALredistribution against general unsecured creditors.19 To this point, theuse of secured credit to redistribute value away from tort claimantsremains unexplored, in spite of the fact that redistribution from tortclaimants is both interesting by itself and provides an ideal test casefor the redistributive theory more generally.

This Article tests the hypothesis that firms facing large tortliabilities will redistribute value away from tort claimants usingsecured debt. As described, tort claimants are the paradigmaticexample of the unsecured creditor in danger of expropriation throughthe use of secured debt. First, the Article examines several differentmethods of identifying companies at high risk for mass tortbankruptcies. Next, it collects and evaluates financial data for thesecompanies and for other publicly traded companies from theCompustat database published by Standard Poor's. Using severalmethods of statistical analysis and controlling for many other factors,the Article compares the amount of secured debt held by high-tortfirms to the amount of secured debt held by otherwise similarcompanies that do not face large tort liabilities.

The results are striking. Companies facing large tort liabilitiesdo not issue abnormal amounts of secured debt. Instead, high-tortfirms appear to issue less secured debt than otherwise similarcompanies not facing bankruptcy. Moreover, these results are robustto many specifications, strongly suggesting that these findings are notthe result of some quirk in the data, but rather are a genuinephenomenon. Individual examples comport with the statistical trends.For example, companies headed for mass tort bankruptcy as a resultof asbestos liability have considerably less secured debt one yearbefore declaring than the average firm within one year of abankruptcy declaration. Large cigarette manufacturers also arelikely candidates to use secured debt to redistribute value away fromtort creditors should tort liabilities bankrupt these firms. And ye t

19. Several empirical finance papers study the priority pattern of lending in light ofeconomic theories of corporate finance. See e.g. Michael J. Barclay Clifford W. Smith, Jr.,The riority Structure o CorporateLiabilities 50 J. FIN. 899, 899 1995); Allen N. BergerGregory F. Udell, Collateral Loan Quality and Bank Risk 25 J. MONETARY ECON. 21, 21 1990). These papers do not consider the role of tort liability or the priority status of tortcreditors in their consideration of loan priorities, however. In an important paper, ProfessorRonald Mann interviewed individuals involved in secured lending and examined their attitudeswith respect to redistribution. Mann, supra note 2, at 630. Mann, however, also does notexamine the implications of tort claimants for the desirability of secured lending.

20. See infra Part IV.

[Vol. 57: 37044


10/45

BANKRUPTCYPhillip Morris, the nation's largest tobacco company, had no secureddebt in the years 2000 and 2001.

The results contradict the redistributional theory for the use ofsecured debt. If redistribution does not occur against nonadjustingtort claimants, it is unlikely to occur in other contexts. The results,however, are consistent with the reduction of agency costsexplanation for secured debt. Tobacco companies and companieswith large asbestos liabilities, for example, are not like other firms infinancial distress. They are successful companies in relatively matureindustries. Firms in industries such as these encounter relatively smallagency costs relative to other firms. As a result, the agency-mitigatingfeatures of secured debt are relatively unattractive to these firms. Iffirms issue secured debt primarily to mitigate agency costs (and notfor redistribution), then these firms should not have large amounts ofsecured debt a prediction confirmed by the findings presented inthis Article.

Indeed, firms with large tort liabilities may be particularly averseto secured debt. A number of observers have noted that securedcreditors have a tendency to prefer liquidation over reorganization inbankruptcy. Mass tort bankruptcy candidates are more likely to beviable firms than other firms in financial distress-their bankruptcy isnot the result of a failing business model, but rather is caused by tort.As a result, reorganization will tend to be more attractive for thesefirms than others. Knowing this, a firm at risk of mass tort bankruptcymay prefer to limit the amount of secured debt to facilitate asuccessful reorganization in bankruptcy.

Several normative recommendations follow from these results.First, superpriority for tort claimants in bankruptcy solves a problemthat is not empirically significant. If firms are not expropriating valuefrom tort claimants under the present priority scheme, then theinefficiencies and inequities decried by advocates for superpriorityare exaggerated. Although there are still justifications forsuperpriority, the benefits of the change must be weighed against thecosts of a change in legal regime. Second, as Professor Scott notes, if

21. See e.g. J. Bradley Johnston, The Bankruptcy Bargain 65 AM. BANKR L J 213, 246 1991). For a judicial recognition of this tendency of secured creditors, see In re Bermec Corp.445 F.2d 367, 369 (2d Cir. 1971) ( We are conscious of the deep concern of the manufacturingsecured creditors lest their security depreciate beyond adequate salvage, but we must balancethat with the Congressional mandate to encourage attempts at corporate reorganization wherethere is a reasonable possibility of success. ).

8] 1045


11/45

DUKE LA W JO URNAnonadjusting creditors are statistically insignificant... then theredistributional claim largely fails and the observed preference ofmany market actors for secured credit is strong evidence of thedominating effects of its cost-reducing properties., Therefore,proposals to change the prevailing regime of full priority for securedclaims on behalf of nonadjusting creditors are empiricallyunsupported. Finally, these results contradict those scholarsadvocating changes in liability regimes because of fears that firmsadjust their capital structure to minimize liability. 3

This Article proceeds as follows: Part I examines the secureddebt debate and provides a theoretical framework for the empiricalresults presented. Part II describes the data collection process andexamines means of identifying high-tort-risk companies. Part IIIpresents statistical analysis of the relation between tort risk andsecured debt usage. Part IV evaluates and interprets the results in thecontext of the theoretical framework presented in Part I.

I. EXPLAINING THE USE OF SECUREDDEBT AND EVALUATING SECURED DEBT'S EFFICIENCY

Firms face a choice between issuing secured and unsecured debt.Unsurprisingly, secured debt has both benefits and costs relative tounsecured debt. Secured debt's priority status in bankruptcy plays animportant role in defining these benefits and costs. 4 The debatedescribed in the Introduction and the empirical analysis presented inPart III examine the relative size and importance of these costs andbenefits. To provide a framework for the empirical examination, thisPart describes the theoretical costs and benefits of secured debt andexamines how different perceptions of these costs and benefits leadsdirectly to varying normative conclusions about the desirability ofsecured debt's priority.

22. Scott, Truth About SecuredFinancing supra note 2 at 46223. Note that many of these conclusions are cautionary in nature. The empirical results

serve more to contradict one theory of secured debt the redistributional theory-than tosupport another theory in particular.

24. These benefits and costs must vary from firm to firm because some firms issue nosecured debt, others issue only secured debt, and a third category issues both secured andunsecured types. See Adler, supra note 2 at 74 (describing the puzzle that secured credit is valuable but is not ubiquitous ). Professor Adler's observation relies heavily on earlier workby Professor Alan Schwartz. Schwartz, Bankruptcy Priorities upra note 2 at 24-25. Thereforeany theory that predicts that secured debt should be either nonexistent or ubiquitous fails thearmchair verification test.

[Vol. 57: 37046


12/45

BANKRUPTCYA The Costs of Secured ebt

Scholars have identified several costs associated with secureddebt with respect to other forms of debt. Professors Bebchuk andFried identify three primary categories of costs: (1) 'contractingcosts'-including the cost of negotiating and perfecting the securityinterest; (2) 'enforcement costs'-the costs of policing the collateral;and, perhaps most importantly, (3) 'opportunity costs'-the costscreated when the security interest prevents the borrower frompursuing efficient activities. 25 Although the secured lender maydirectly bear many of these costs, the debtor must pay a higherinterest rate to compensate the lender for these expenses.

Professor Ronald Mann's investigation suggests thatenforcement costs are considerably higher for secured loans than forunsecured loans.26 Secured lenders must monitor their securities toensure that their rights are protected. This monitoring is costly.27

The largest cost associated with secured debt relative tounsecured is the prevention of efficient investment activities by thedebtor. A secured lender may prevent a debtor from using itscollateral in a productive investment if it perceives that theinvestment may reduce the lender's recovery (even if the investmentis profitable on average). The lender's goal is to maximize its ownrecovery, and not the total value of the firm. Renegotiation of asecured loan may ameliorate this problem, but the renegotiation isitself costly. As one borrower quoted by Professor Mann stated abouta secured loan, You just don't have the same flexibility of dealingwith your properties as if you owned them unencumbered. ' This lossof flexibility is costly for the secured debt issuer.

Secured credit may also increase the cost of reorganization inbankruptcy. Because secured creditors have priority, they tend toprefer low-risk bankruptcy strategies such as liquidation, even ifliquidation destroys value. 29 Furthermore, secured debt may increase

25 Bebchuk Fried, Uneasy Case I supra note 2, at 877.26 See Mann, supra note 2, at 663 (noting that information costs strongly encourage

unsecured credit in transactions involving large borrowers ).27. See id. at 663-64. Mann notes that these costs will only be incurred because there is a

benefit in increased repayment probabilities. Id at 663. Part I.B.2 examines these benefits.28. See id. at 665 (quoting Telephone Interview, Joseph W. Robertson, Jr., Chief Financial

Officer, Weingarten Realty Investors (July 11, 1995)).29. See Arturo Bris, Ivo Welch Ning Zhu, The Costs of Bankruptcy: Chapter

Liquidation vs Chapter Reorganization 4 (Int'l Center for Fin. at Yale Sch. of Mgmt.,

8] 1047


13/45

DUKE LA W JO URNA Lfree-rider problems. The firm's bundle of assets may be worth moretogether than apart. Each secured creditor may attempt to extract thisvalue from other creditors by attempting to hinder the attemptedreorganization. Because the secured creditor with priority loses littleif negotiations fail and end in liquidation and has a right that is clearlyassociated with a particular asset, the secured creditor is uniquelyplaced to hold up the negotiations and demand excess value.3 As aresult, large amounts of secured debt may obstruct efficientbankruptcy reorganizations-an important additional cost created bysecured as opposed to unsecured debt.

Finally, the availability of security with a priority claim inbankruptcy raises a firm's cost of lending more generally. Unsecuredcreditors will demand costly covenants restricting later secured debtor an interest rate premium to protect themselves against the risk thatthe issuance of later secured debt may dilute the value of theirclaims. In both cases, a firm must pay a higher price for unsecuredcredit.B. The Benefits of Secured Debt

Given the costs described in Section A, secured debt must offeroffsetting benefits if debtors are ever to issue security. Unfortunately,there is less consensus about the benefits of secured debt to firmsthan there is regarding the costs. This Section details some of themany benefits proposed by scholars as explanations for the existenceof secured debt.

1 Controlling Monitoringand Agency Costs. Most attempts todefine the benefits of secured debt focus on the ways in whichsecured credit can better control agency costs within the firm byreducing conflicts of interest between the debtor's managers(representing the residual equity claimants) and the firm's debtholders. 32 Better control of agency costs reduces the interest rate a

Research Paper Series, Working Paper No. 04-13, 2004), available athttp://ssrn.com/abstract=523562 (stating that when banks are secured creditors, they prefer firmsto be liquidated and do not favor reorganizations).

30. See Patrick Bolton & David S. Scharfstein, OptimalDebt Structure and the Number ofCreditors 104 J. POL. ECON 1, 2 (1996).

31. This is part of what Professor Alan Schwartz has termed the puzzle of secured debt.See Schwartz, Taking the Analysis o Security Seriously supranote 2, at 2079-80.

32. See Scott, Truth About Secured Financing upranote 2, at 1448.

[Vol. 57: 37048


14/45

2008] BANKRUPTCY 1049firm must pay; the more confident the lender is that the firm will notbehave opportunistically, the less of a premium the lender willdemand.33Secured lending appears to offer a partial solution to the overinvestment or risk-alteration problem.' Security enables alender to exact harsh penalties on a debtor in the event of default. Forexample, the secured lender may foreclose on potentially criticalassets. Thus, if a borrower attempts to engage in covenant-violatingrisk alteration, a secured lender has considerably more power than anunsecured lender to prevent opportunistic risk alteration.35 Becauserisk alteration may be inefficient, the leverage gained by the securedcreditor may enable socially preferred outcomes.

Secured lending may also mitigate the underinvestmentproblem.36 The underinvestment problem occurs when a firm choosesto reduce its investment in a project because it must share the benefitsof the project with a creditor/joint venturer. Instead, the firm mayprefer to invest in a less promising project in which the firm retains allthe profits. Security enables creditors/joint venturers to preventunderinvestment. If a firm violates a covenant associated with alending contract in a way that will reduce the creditor's return, thenthe secured creditor can threaten to foreclose on the collateral. This

33. This discussion focuses on secured debt as a means of reducing risk alteration. Otheragency costs related to theories of secured debt focus on secured debt's ability to decreasemonitoring costs. ee DOUGLAS G BAIRD THOMAS H. JACKSON, CASES, PROBLEMS, ANDMATERIALS ON SECURITY INTERESTS IN PERSONAL PROPERTY 324-28 1987); Buckley, supranote 2, at 1396; Jackson Kronman, supra note 2, at 1150-51; Levmore, supra note 2, at 49.

34. For analyses of this issue, see Elazar Berkovitch E. Han Kim, FinancialContractingand Leverage Induced Over- and Under-Investment Incentives 45 J FIN. 765, 765 1994); Bolton Scharfstein, supra note 30, at 2; Scott, Relational Theory supra note 2, at 901; Charles W .Smith Jerold S. Warner, On FinancialContracting:An Analysis of Bond Covenants 7 J. FIN.ECON. 117, 117 (1979). Another prominent explanation of secured debt (and priority rules moregenerally) on the basis of agency costs and primarily risk alteration-controlling properties isKanda Levmore, supra note 2, at 2106.

35. Many scholars list specific ways in which the security can reduce risk alteration. Seee.g. Carlson, supra note 2, at 1637; Rene M. Stulz Herb Johnson, An Analysis of SecuredDebt 14 J. FIN. ECON. 501, 502 1985); Triantis, A Free-Cash-Flow Theory supra note 2, at2157. For a list of means of risk alteration, see Bebchuk Fried, Uneasy Case 1 supranote 2, at876.

36. Professor Stewart Myers offered the original explication of the underinvestmentproblem. See generally Stewart C. Myers, Determinantsof CorporateBorrowing 5 J. FIN E ON147, 147 1977) (examining the gap in modern finance theory on the issue of corporate debtpolicy ). The underinvestment problem was related to secured debt by Berkovitch Kim,supra note 34, at 765-66, and Smith Warner, supranote 34, at 119.


15/45

DUKE LA W JOURNAL [Vol. 57:1037harsh penalty deters the firm from inefficiently starving the jointventure of funds.

Similarly secured lending may reduce the number of opportunistic defaults. 7 An opportunistic debtor will declarebankruptcy when there is no shortage of liquidity if it believes that abankruptcy reorganization will improve its credit terms. In response,a lender must demand potentially excessive interest rate premiums tocompensate for the strategic default risk. Secured debt may offeranother solution. Because secured debt makes bankruptcyeor niztionmor cosly d 8reorganization more costly and uncertain, it will deter opportunisticbankruptcy declarations. When opportunistic bankruptcies are a largerisk, the secured debt solution may be more attractive. Whenopportunistic bankruptcies are unlikely, however, secured debt willbe less attractive because it will raise the cost of potentially efficientbankruptcy reorganizations.39

37. This discussion borrows from the work of Bolton Scharfstein, supra note 30 at 2.38. See supranote 29-30 and accompanying text.39. In their important article, Professors Bebchuk and Fried argue that these benefits of

secured debt are independent of secured debt's priority. Instead, they argue that the benefitsarise from the special rights accorded to secured creditors outside of bankruptcy. BebchukFried, Uneasy Case 1 supra note 2 at 875. Indeed, the unique characteristics of secured debt,such as the right to foreclose on an asset in response to default, are related to state law rights offoreclosure and not to the priority of security in bankruptcy. Bebchuk and Fried's claim that theagency cost-reduction abilities of secured debt is independent of bankruptcy priority isincomplete, however. Bebchuk and Fried correctly note that foreclosure is a state law right thatenables a creditor to effectively deter opportunistic behavior (such as risk alteration) by thedebtor. Id They fail to discern, however, that the effectiveness of the foreclosure deterrentdepends critically on the priority that secured credit receives in bankruptcy. The argument runsas follows: Suppose that secured creditors do not enjoy priority in bankruptcy. Suppose furtherthat a debtor attempts to violate the terms of a secured debt contract and engages inopportunistic risk-altering behavior. The secured creditor threatens to foreclose, seeking todissuade the debtor from engaging in risk alteration. In turn, the debtor threatens to declarebankruptcy. If the secured debtor does not enjoy priority in bankruptcy, then the debtor'scounterthreat is effective. The secured creditor will be reluctant to foreclose because it riskslosing value while sharing priority with other creditors. The state law right of foreclosure istoothless if secured creditors do not enjoy priority in bankruptcy. When the secured creditorsenjoy priority, however, the debtor's bankruptcy threat does not intimidate the secured creditor,which can be confident that it will recover its loan (along with interest) because it enjoys priorityin bankruptcy. Indeed, secured lenders are more likely to place a creditor in bankruptcy thanother lenders, suggesting that they are far less afraid of bankruptcy, and therefore better able toexercise leverage, than other creditors.

1050


16/45

BANKRUPTCY2. Redistribution-Priority Related. Redistribution offersanother explanation for the use of secured debt. The redistributional

capacities of secured debt operate as follows: First, suppose that afirm has nonadjusting creditors. Nonadjusting creditors are creditorswho do not adjust the interest rate charged to the firm in response toevents that alter the expected recovery of the loan should bankruptcyoccur. Next, suppose that the firm needs to issue debt and mustchoose between secured and unsecured debt. If the firm issuessecured debt, then the new creditor will demand a lower interest rateas compared with unsecured debt because the new creditor will bemore confident of recovery in bankruptcy due to secured debt'spriority. Secured debt reduces the value of nonadjusting creditors'claims, however. Because the new creditor's claims are secured, thenonadjusting creditors will only recover in bankruptcy after thesecured creditors are paid in full. Thus, more secured credit meanslower recoveries for other creditors. In response to this effect,preexisting creditors should charge a higher interest rate tocompensate for the increased risk of nonrecovery in bankruptcy ordemand covenants to limit the probability of redistribution. For avariety of reasons, nonadjusting creditors do not do so. As a result,the firm can issue new secured debt and obtain a lower interest rate,thereby redistributing value away from the nonadjusting creditorswhose claims have been diluted through the use of security.43Secured debt's redistributional benefits to a firm stem from theexistence of nonadjusting creditors. Scholars have identified severalgroups of nonadjusting creditors. These include (1) privateinvoluntary creditors such as tort claimants, 2) government tax andregulatory claims, 3) voluntary creditors with small claims, and 4)prior voluntary creditors. The third and fourth classes ofnonadjusting creditors may adjust their interest rate to reflect

40. See e.g. Schwartz, Bankruptcy Priorities supra note 2, at 30-33; James H. Scott,Bankruptcy Secured Debt and Optimal Capital Structure 32 J IN. 1, 1 (1977); James H. Scott,Bankruptcy Secured Debt and Optimal CapitalStructure: Reply 34 J. FIN. 253, 253 1979). Seegenerally Bebchuk Fried, Uneasy Case 1 supra note 2 (presenting a new and more generalanalysis of the redistributional theory).

41. See e.g. Christopher W . Frost, Asset Securitizationand Corporate Risk Allocation 72TUL. L. REV. 101, 125-26 1997); Schwartz, supranote 2, at 7-8.

42. See Bebchuk Fried, Uneasy Case 1 upra note 2, at 885-86.43. See id. at 864-66.44. See id. at 882-91.45. Id.

2 8] 5


17/45

DUKE L W JOURN L

nonadjustment. The first two classes cannot. Of these classes, themost frequently noted and commented upon are tort claimants.46Tort claimants become creditors when firms do not carry enoughinsurance to cover all tort claims. Insurance may not cover all tortclaims because insurers insist on a coverage limit or because firmshave an incentive to underinsure. 7 Whatever the cause, when firmsare underinsured, tort victims of the firm must recover from the firmrather than the firm's insurer. The size of the tort claimant's claim isfixed by the size of damages. 8 The tort claimant is therefore theparadigmatic nonadjusting creditor-the tort claimant cannot claimadditional compensation if a later secured loan reduces the expected

tort recovery in bankruptcy. Moreover, tort claimants, as involuntarycreditors, enjoy no contractual protections against redistribution suchas covenants. As a result, tort claimants are highly exposed toredistribution. A firm with many tort creditors should pay aconsiderably lower interest rate for secured debt, which enjoyspriority over the tort creditors, as opposed to unsecured debt, whichmust share firm assets pro rata with the tort claimants. Theredistributional theory suggests that firms should therefore issuesecured debt. Tort claimants will suffer from the secured debt, as thetort claimant s bankruptcy share is reduced without anycountervailing increase in payoffs.

46. See e.g. id. at 882 (presenting tort claimants as the first type of creditor in theirtaxonomy of nonadjusting creditors); LoPucki, supra note 2, at 1898-99; Scott, BankruptcySecured Debt nd Optimal Capital Structure supra note 40 at 2-3; Shupack, supra note 2, at1094-95.

47. See Bebchuk Fried, Uneasy Case 1 supranote 2, at 882; Henry Hansmann ReinierKraakman, Toward Unlimited Shareholder Liability for Corporate Torts 100 YALE L.J. 1879,1889 1991).

48. This is especially true for late-arriving tort claimants whose claims arise in or nearbankruptcy, precluding the possibility of compensation for nonadjustment through higherinterest rates.

49. Because fraudulent conveyance doctrines in bankruptcy only apply to transactionsmade within two years of filing and exempt transactions for which reasonably equivalentvalue was paid, fraudulent conveyance offers limited protection for tort claimants againstredistributional secured debt. See 11 U.S.C. 548(a) (2000 Supp. V 2005).

50. The other classes of nonadjusting creditors may suffer from similar forms ofredistribution. These nonadjusting creditors may demand interest rate premiums to compensatefor potential redistribution or they may demand contractual protections instead. Although theremay still be some inefficiencies associated with redistributional secured debt in these contexts,they are much less problematic than the case of involuntary creditors. Redistribution does notnecessarily imply that a firm's incentives to take precautions against torts will be reduced. For adiscussion, see Yeon-Koo Che Kathryn E. Spier, StrategicJudgment Proofing 14-15 (The

[Vol. 57:1037052


18/45

BANKRUPTCYC The Normative Implicationsof Theories of Secured Debt or

BankruptcyPrioritiesWhether or not secured debt should enjoy priority in bankruptcydepends upon the relative predictive accuracy of the two theories of

secured debt presented in Section B. If the agency cost reductiontheory is primarily correct in describing secured debt usage, thensecured debt should receive priority over other types of debt. Priorityincreases the leverage the secured lender enjoys over the creditor. Inturn, this leverage enables the secured lender to reduce potentiallyinefficient opportunistic behavior by the debtor. Any reduction inpriority for secured credit may reduce leverage and inhibit theusefulness of secured debt as an agency cost-reducing tool.

If one finds that firms issue secured debt for redistributionalreasons, by contrast, then priority for secured debt is less desirable.Redistribution leads to inefficiencies. Redistribution involves thetransfer of value from nonadjusting creditors to newly arrivingsecured creditors. No value is created. Instead, value is shuffled. Ifsecured debt costs more to issue than unsecured debt (for example, iftransactions costs are higher in the case of secured loans), thensecured loans would consume valuable resources for a benefit,redistribution, that transfers but does not create wealth-aninefficient outcome. Secured debt issued for redistributional reasonsalso may lead to other inefficiencies. In the tort context, for example,redistribution may lead to excessively low levels of precaution.

Suppose a firm with no insurance commits a large tort thatthreatens to force the firm into bankruptcy. If the firm cannot issuedebt with priority over tort claimants, then it will be forced to borrowat high rates; contract creditors will know that they will be forced toshare pro rata with the large group of tort creditors in the event ofbankruptcy and will demand to be compensated accordingly. In thissituation, the firm will take precautions to avoid mass torts, whichraise the firm's cost of credit.Now suppose, however, that secured debt enjoys priority overtort claimants and that firms issue secured debt for redistributionalreasons. In this scenario, the firm's incentives for precaution are

Center for the Study of Indus. Org. at Northwestern U. Working Paper No. 0081 2006 ,availableat http://www.wcas.northwestern.edu/csio/Papers/2006/CSIO-WP-0081.pdf.

51. Of course, elements of both theories may be correct. The scholarly debate concernsthe relative importance of the explanations in determining the real-world usage of secureddebt.

8] 1053


19/45

DUKE LA W JOURNAgreatly reduced. If a tort occurred, the firm's cost of credit would notbe greatly altered. Instead, the firm would issue secured debt.Contract creditors would not demand high interest rate premiumsbecause the secured debt enjoys priority in bankruptcy and will befully repaid before the large class of tort claimants receives anything.Thus, secured debt enables a firm to avoid the full costs of largetorts.52 In response, a firm will have less of an incentive to takeprecautions to avoid the large tort.

If redistribution is the primary motive for secured debt issuance,then priority for secured debt also leads to inequitable distributions.Whereas unsecured contract creditors are voluntarily unsecured andmay receive compensation for their lack of security in the form ofhigher interest rates, tort claimants can not choose their security level.They are unwillingly exposed to redistribution. Furthermore, tortvictims may have extraordinary liquidity needs that go unmet as aresult of secured debt's priority status. 3 In response to theseefficiency and equity concerns, many adherents of the redistributionaltheory have advocated a reduction in secured debt's priority.Arguments to grant tort claimants priority over other creditors have along history. One commentator describes the argument as follows:

One effective way of deterring insolvency and encouraging optimalprecaution levels would be to alter the priority scheme by giving tortcreditors superpriority status. Under this system of superpriority,tort creditors would be paid before all priority creditors, securedcreditors, and unsecured creditors. Since secured creditors wouldnot be guaranteed payment at the head of the line in this regime,they would bear a portion of the risk of insolvency. Securedcreditors would be forced to price risk into credit and would in turnforce firms to internalize this risk through credit prices thatcorrespond to precaution levels.... To force firms to include the fullcost of accidents in business decisions, it is crucial to prioritize tortclaims above all other priority claimants, even secured creditors. Iftort debt were given priority, all creditors would have an incentive tomonitor business risk-taking. Because they are often less diffuse and

52. In this context, secured debt also diminishes a firm's incentives to purchase insurance.If a firm can use secured debt to avoid making payments to tort creditors and thereby diminishthe amount ceded by equity claimants to tort claimants, there is less of a need to purchaseinsurance that covers the firm's obligations to tort claimants.

53. See Yair Listokin & Kenneth Ayotte, Protecting Future Claimants in Mass TortBankruptcies 98 Nw. U. L. REV. 1435, 1438 2004).

54. See supranote 11 and accompanying text.

[Vol. 57: 37054


20/45

BANKRUPTCYmore savvy than shareholders, creditors are in the best position tomonitor levels of risky activity.55These arguments apply only to noninsured tort liabilities. Whentort liabilities are insured, the insurer has an incentive to price the

insurance appropriately and encourage the firm to take appropriateprecautions. Furthermore, tort claimants do not suffer from priorityfor secured debt when firms are insured because the tort claimantsreceive compensation from the insurance company and not thetortfeasing firm.

Superpriority for tort claimants is not the only proposedbankruptcy priority reform. Noting that tort claimants are not theonly nonadjusting creditors, Professors Bebchuk and Fried advocate a partial priority rule for secured debt. 6 They believe that partialpriority would reduce the amount of secured debt issued forinefficient redistributional reasons. Furthermore, Professor LynnLoPucki points to redistributional secured debt as an example of agrowing trend leading toward the death of tort liability.Concluding that tort claimant priority alone is insufficient in the faceof this trend, Professor LoPucki considers unlimited shareholderliability and consensual creditor liability as potential solutions, butconcludes that even these radical steps are potentially inadequate. 9

The debate over the primary reasons for secured debt's usageinvolves high stakes. If secured debt is primarily used to reduceagency costs, then it is efficient for secured debt to enjoy priority overother debt. If secured debt's primary use is to redistribute moneyfrom one class of creditors to another, however, then a host ofchanges to the bankruptcy priority rules are warranted.

II. EMPIRICAL ANALYSIS: SECURED DEBT AND TORT LIABILITYThe discussion in Part I described the importance of

distinguishing between the agency-cost theory and theredistributional theory of secured debt. This Part presents anempirical framework for distinguishing between the two theories by

55 See Note, supranote 11 at 2562.56. Bebchuk Fried, Uneasy Case 1 supra note 2, at 866. Partial priority rules treat a

portion of all secured debt as unsecured debt.57. d58. LoPucki, supra note 17, at 14-23.59. Id at 63.

8] 55


21/45

DUKE L W JOURN Lfocusing on secured debt usage in response to noninsured tortliabilities.A Theoriesof Secured Debt and Predictions egarding Secured DebtUsage and Tort Liability

The redistributional theory predicts that firms with large,noninsured tort liabilities should have large amounts of secureddebt,' and recall that tort claimants are uniquely exposed toredistribution.61 They do not enjoy contractual protections such ascovenants, nor do they voluntarily choose to forego such protection inexchange for greater compensation. As a result, firms faced with largetort liabilities have particularly strong motives to engage inredistribution. The spread in interest rates between secured andunsecured debt, and thus the payoff for engaging in redistributionthrough secured debt, should be particularly high for high-tortfirms. ( High-tort firms are those with large actual or potential tortliabilities, such as asbestos companies, tobacco companies, or nuclearpower plant operators.) Moreover, high-tort firms will not have toengage in the hassle of renegotiating covenants with the large class ofunsecured tort claimants. Thus, redistribution is at its easiest andmost compelling for high-tort firms. If redistributional motives everdrive secured debt issuance, then firms facing noninsurable tortliabilities should engage in large amounts of secured borrowingrelative to otherwise similar firms without large tort liabilities.By contrast, the agency-cost theories of secured debt do no tpredict that high-tort firms should issue large amounts of secureddebt. Firms facing large tort liabilities are no more susceptible to riskalteration or other agency costs than other firms. Indeed, high-tortfirms may be less susceptible to these agency costs than an averagefirms. High-tort firms' risks stem primarily from partially exogenous,nonperformance-related factors such as court decisions.62 Other firms,by contrast, encounter more significant performance-related risk thatcan be adjusted through risk shifting and controlled with secureddebt. Furthermore, many of the high-tort firms identified in thisArticle are in mature industries with stable cash flows, making them

60. See supra Part I.B.2.61. See supr Part I.62. Indeed, the tort risks often derive from decisions made by previous executives (such as

in the case of asbestos and tobacco) and therefore have little if anything to do with the firm scontemporaneous operating performance.

[Vol. 57: 3756


22/45

BANKRUPTCYpoor candidates for secured debt given secured debt s undeniablyhigh costs.63

Thus, high-tort firms offer a compelling empirical test of theredistributional theory. If high-tort firms do not have large amountsof secured debt, it is unlikely that they use secured debt for large-scale redistribution.B. Identifying Firms with Large Tort Liabilities

This Article adopts several techniques for identifying firms withlarge, noninsurable tort liabilities. First, this study focuses on firms inindustries with well-known, large mass tort liabilities. Firms with largeasbestos liabilities are the most prominent example. To identify thesefirms, this Article draws upon the research of Professor MichelleWhite, who compiled a list of companies that declared bankruptcy asa result of asbestos liability.6 Note that many of these firms are notasbestos manufacturers; most manufacturers declared bankruptcywell before the time period examined in this sample.65 Instead, thefirms come from a number of industries and often became exposed toasbestos liability by acquiring firms with some early and little-notedconnection to the asbestos industry.6 These firms are identifiedthrough the use of a dummy variable that equals one if the firm ispresent on Professor White s list and zero otherwise.Tobacco firms are another widely publicized group of firms withpotentially large mass tort liabilities. Since the 1990s, tobaccocompanies have become the target of suits from numerous sources,including smokers, secondhand smokers, and representatives of states

63 Although firms in mature industries are unlikely to default on their loans, they may belikely to waste their free cash flow on unproductive projects. As a result, high debt loads formature companies may be efficient ways to constrain management from wasting the free cashflows. See Michael C Jensen, Agency Costs Of Free Cash Flow, Corporate Finance, andTakeovers, 76 AM. ECON REV. 323, 323 1986). Because risk alteration is not an importantconcern for such companies, however, the need for secured debt, as opposed to other forms ofdebt, is limited.

64. See Michelle J White, hy the Asbestos Genie Won t Stay in the Bankruptcy Bottle, 70U. CIN. L. REV. 1319, 1320 2003).

65. For example, the Manville Corporation, the largest asbestos manufacturer, declaredbankruptcy in 1982. In re Johns-Manville Corp., 36 B.R. 743 (Bankr. S.D.N.Y. 1984).

66. See, e.g. Note, Successor Liability,Mass Tort, and Mandatory-LitigationClassAction,118 HARV. L. REV. 2357, 2374-77 2005) (discussing G F Corporation s difficulties as a resultof asbestos liability assumed by acquiring another company that had made asbestos yearsearlier).

2008] 1057


23/45

DUKE LA W JO URNALand the federal government. 7 Although these suits have yet to forcethese firms into bankruptcy, there is a high mass tort bankruptcy riskfor tobacco firms. State tobacco bond interest rates provide evidenceof this risk. Adverse tort verdicts have made states reluctant to issuetobacco bonds securitizing settlement payments from tobaccocompanies.69 Because investors fear that tort liability will forcetobacco companies into bankruptcy, tobacco bonds carry a highinterest rate premium.

Tort risk is a salient characteristic of both tobacco and asbestosfirms, and it should surprise no one if these firms adopt capitalstructures with this risk in mind. To broaden the sample of high-tortfirms beyond these two industries, however, a more rigorous methodof identifying high-tort industries is necessary. This study uses theWestlaw databases LLST TES and LLFEDS to assist in thistask. The procedure for identifying high-tort-risk industries operatesas follows: First, the official census bureau titles of industries-knownas the Standard Industrial Classification (SIC) titles-were obtainedfrom the U.S. Census Bureau.7 Next, the key phrase7 for each three-digit SIC industry group code title was typed into Westlaw along withthe following command:

[Industry Name] /s products liabilityThe resulting hits were recorded. This search gives an indication ofthe number of times an industry is associated with products liabilityrisk the cause of almost all the mass tort bankruptcies heretofore.

67. For a discussion of many of these suits, see David Hechler, Tobacco Takes It on theChin NAT L L.J., Feb. 19, 2003, http://www.law.com/jsp/article.jsp?id=1043457953024.

68. Tobacco companies have large obligations to states stemming from l rge settlementof a 1998 class action lawsuit against tobacco companies by many states' attorneys general. Toplug large deficits, some states have securitized their receipts from tobacco companies. Thesebonds are called state tobacco bonds. See Al Baker Jonathan Fuerbringer, Shift in BondsHas States Rethinking Tobacco Plans N.Y. TIMES, Apr. 10, 2003, at A18.

69. Id.70. The Census Bureau gives each industry an official SIC title as well as an official three-

digit SIC Code. For a list of these industrial titles and codes, see SIC Division Structure,http://listsareus.combusiness-sic-codes.htm (last visited Dec. 7, 2007).

71. Many of the SIC titles contain extraneous words. For example, the paper industry is nottermed the Paper industry but rather the Paper and Allied Products industry. For thisstudy, the key term of each SIC title was used. For example, searches were conducted using theword Paper and not Paper and Allied Products. list of these phrases is available from theauthor upon request.

1058 [Vol. 57:1037


24/45

BANKRUPTCYTo control for the fact that some industries' names are more commonthan others, I conducted another control search using the command

[Industry Name] /s breach of contractand I recorded the resulting hits. Industries that generated a lot ofhits for the first search simply because they had a common nameshould also have large number of hits on the control search. Bycontrast, industries with genuinely high products liability risk shouldhave many more hits on the first search than on the second. Theindustries with large numbers of hits for search but not for search 2were identified as high-tort industries with a dummy variable equal to

72one.Although this procedure is imperfect, it provides a rough-and-ready objective means of identifying industries with high tort risk.Furthermore, the results from this procedure are intuitively

reasonable. The two highest scoring industries according to thismethod were the tobacco and asbestos industries. Other industriesidentified as high tort using the Westlaw method includepharmaceutical firms, surgical and medical equipment makers, paintmanufacturers, pesticide makers, tire manufacturers, small armsmanufacturers, household appliance manufacturers, and toy andsporting goods equipment makers. Note that the number ofpharmaceutical and medical equipment firms dwarfs the number ofother firms. As a result, this study will treat pharmaceutical andmedical equipment firms separately from the other group of high-tortfirms.C FinancialDataand Summary Statistics

This study focuses on manufacturing firms with SIC codesbetween 2000 and 4000 This range includes only manufacturing firmsand excludes financial and services firms. Financial data for all thefirms in the sample was collected from Compustat, a proprietarydatabase containing detailed financial information for publicly tradedfirms traded on American stock exchanges.7 Table presentssummary statistics for many of the important variables used in the

72 Large was defined as twice as many hits for search than for search 273 Because Compustat focuses on public firms, the conclusions are necessarily speculative

with respect to closely held firms. Given the strength of the redistributional motive for high-tort-risk firms, however, a failure of the redistributional prediction for publicly traded firmsbodes ill for the theory's predictive power with respect to closely held firms.

8] 59


25/45

DUKE LAW JOURNAL [Vol. 57:1037analysis. The dataset from which these figures are based contains datafrom 5,592 firms.74 For each firm, the dataset contains an average ofalmost seven years of data (6.802 years per firm on average). Thetotal number of firm-year observations in the dataset is therefore5,592*6.802 38,040 firm years. Table 1, as well as all of the otherincluded tables, uses the firm (and not the firm year) as the basic unitof observation. Thus, the averages in Table 1 are averages of theaverage value for each firm.Table 1. Summary Statistics

Percent of Number ofStandard Observations Observations

Variable ean eviation at Zero Firms)Secured ebt(milion of 36.8 270.3 18.4% 5,581millions of dollars)Total Long-Term Debt 219.2 1,329.7 8.6% 5,580

(millions of dollars)Total Firm Assets

(millions of dollars) 1,065.5 5,951.9 0.0% 5,581Property, Plant, and

Equipment-Total 680.7 4,154.9 0.0 5,581(millions of dollars)

Employees 4,651.2 20,413.6 0.0% 5,425Secured Debt-to-Total 0.31 0.28 18.4% 5,581

Debt RatioSecured Debt-to-Hard 0.18 0.24 18.4% 5,581

Assets RatioSecured Debt-to-Total 0.09 0.13 18.4% 5,581

Assets Ratio

74. This includes pharmaceutical and medical firms.75. To explain further, suppose that Firm A is a representative firm, and there are two

years (1 and 2) and two variables (secured debt and tort risk). Suppose that in year 1 Firm A has1 unit of secured debt and 1 unit of tort risk and that in year 2 Firm A has 1.5 units of secureddebt and 1 unit of tort risk. Thus, Firm A's average secured debt is 1.25 1+1.5)12, and FirmA's average tort risk is I 1+1)/2. In calculating the averages presented in Table 1, Firm A isentered as one observation with secured debt equal to 1.25 and tort risk equal to 1, rather thanas two firm-year observations. The reasons for this choice are explained in Part III. See infratext accompanying notes 87-88.

1060


26/45

B NKRUPTCYCompustat collects data for large, publicly traded firms. As a

result, the average firm in the dataset is large. It has 36.8 million ofsecured debt, 219.2 million of total long-term debt, hard assets ofover 680 million, total assets of approximately 1.06 billion, andmore than 4,650 employees. The data for the firms is widelydispersed. Note how the standard deviations for many of the variablesin Table 1 are greater than the means. This occurs because the sizedistribution of the firms in the sample is uneven. There are manymore smaller-than-average firms in the sample than larger-than-average firms. A few very large firms help skew the distributions ofthe variables 6

Not all firms have secured debt. Indeed, approximately 18percent of the firms in the sample have no secured debt at all. Thisnumber is similar to the figure obtained from previous research. Onaverage, secured debt makes up 31 percent of all debt for firms thathave at least some debt (the secured debt-to-total debt ratio is .31 .Thus, secured debt is an important part of firms' debt composition,although it is far from ubiquitous. Creditors do not hold collateral onmost of the firms' assets. In fact, the average firm in the sample has aratio of secured debt to property, plant, and equipment (a term thatreflects the number of hard assets the corporation has) of .18-only18 percent of the average firm's hard assets are securitized, suggestingthat lack of hard assets does not constrain most firms from issuingsecured debt. Secured debt comprises an even smaller percentage offirms' total assets; the average secured debt-to-total assets ratio isonly .09. Note that the secured debt variable includes capitalizedleases. Thus, the secured debt variable captures any attempt bycompanies to evade tort creditors through long-term sale andleaseback arrangements or similar types of securitizations.

76. Note that the values of most variables used infra are capped above and below at theand 95 percentiles, respectively. This technique helps prevent outliers from driving the results.This process does not affect the primary focus of this Article (the use of secured debt by high-tort-risk firms). It does affect the estimated coefficients of many of the control variablespresented in Table 3, however.

77 See Barclay Smith, supra note 19, at 904 tbl.II (finding that 24 percent of firms didnot have secured debt).

78 Note that some firms have no debt of any type. These firms do not appear in thecalculation of the ratio of secured debt to total debt because the ratio is undefined for thesefirms-the denominator, total debt, is zero. This explains why there are fewer firm-yearobservations making up the secured debt-to-total debt ratio as compared with the other ratiospresented in Table 1. It also explains why the secured debt-to-total debt ratio does not simplyequal the mean for secured debt amount divided by the mean for total debt amount.

8] 1061


27/45

DUKE LAW JOURNALIII. STATISTICAL ANALYSIS OF THE

RELATIONSHIP BETWEEN TORT RISK AND SECURED DEBT USAGEWith these stylized facts regarding secured debt usage in mind,

examine the relationship between secured debt usage and tort risk,and this Part presents the data analysis. It begins with an analysis ofsummary statistics of firms with different tort characteristics. The Partcontinues with Tobit analyses, a form of regression analysis to helpcontrol for other factors that might be driving any association ofsecured debt usage with tort risk. Finally, the Part relies on time-series evidence to isolate further the impact of tort risk on secureddebt usage from other industry specific characteristics driving secureddebt usage.A Simple StatisticalAnalysis of Secured Debt Usage

Recall that the redistributional theory of secured debt predictsthat firms at risk of tort liability-related bankruptcies should havegreater amounts of secured debt relative to otherwise similar firmsthat do not have high tort liabilities. Before testing this hypothesisthrough regression analysis, this Section presents some simplestatistical comparisons of high-tort firms with respect to otherfirms.

Table 2 divides the firms in the sample according to severalmeasures of tort risk and compares the ratios of secured debt to hardassets and secured debt to total debt for high-tort-risk and low-tort-risk firms. These ratios are chosen over other measures of secureddebt usage for several reasons. First, the theoretical literatureexamines the tradeoff between secured debt and other types of debt.This tradeoff is best analyzed empirically through the use of thesecured debt-to-total debt ratio. Second, using a ratio facilitatesempirical comparisons of firms of very different sizes by puttingrelative secured debt usage of any firm on a comparable scale.The first row of Table 2 presents figures for firms that are notidentified as high tort risks. This group consists of all firms notidentified as high tort using the methods described in Part II.Secured debt accounts for 31 percent of total debt for these non-high-tort control group firms, and only 16 percent of the control groupfirm's hard assets are secured. The low figure suggests that limits onsecurable assets do not constrain non-high-tort firms from obtainingmore secured debt.

[Vol 57: 3762


28/45


29/45

UKE L W JOURN

between cigarette firms and the control group may be the cause ofthese differences. One possible explanation of these differenceswould be that cigarette firms' secured debt capacities are limited bytheir securable assets. If so, cigarette firms are unable to obtain moresecured debt because they have nothing to use as security. The data,however, do not support this hypothesis. Cigarette firms have a lowerratio of secured debt to hard assets than the control group 12 percentas opposed to 18 percent). Moreover, both groups appear to haveplenty of securable assets available to collateralize loans; less than 20percent of hard assets are secured.

These results repeat themselves for firms facing large asbestosliabilities. Firms facing large asbestos liabilities are ideally situated touse secured debt for redistribution. Nevertheless, secured debtaccounts for a smaller proportion of their debt loads than for thecontrol group 15 percent for high asbestos liability firms as comparedto 31 percent for the control group). Here too, the results cannot beattributed to lack of securable assets. Secured debt equals only 8percent of these firms' hard assets, making it unlikely that the assetconstraint is preventing these firms from obtaining more secureddebt.

As with cigarettes, however, the results must be treated withcaution because of the limited size of the asbestos firm sample. Toaddress this issue, Table 2 presents a pooled high-risk sampleconsisting of tobacco and asbestos firms as well as two other highprofile mass tort bankruptcy firms (Dow Corning and A.H. Robins).The pooled sample also has considerably lower ratios of secured debtto total debt 12 percent) and secured debt to hard assets 13 percent)than the control group. Firms in the Westlaw-identified high-tortindustries' have considerably lower secured debt proportions thanthe control group of non-high-tort firms.

The results for pharmaceutical and medical devices firms providelimited support for the redistributional theory. Indeed, these firmshave a slightly higher ratio of secured debt to total debt than thecontrol group 33 percent for the pharmaceutical and medical devicesfirms as opposed to 31 percent for other firms). These results shouldbe treated with caution for at least two reasons, however. First, thelink between the pharmaceutical industry and tort risk is far weaker

84. ee supra Part II.B.

1064 [Vol. 57: 37


30/45

BANKRUPTCYthan for asbestos or tobacco. Second, many other factors may becausing these differences.

Because the simple statistics presented here do not control for amyriad of other factors, firm conclusions based on these results areimpossible. Nevertheless, one observation is justified: redistribution'sability to explain the pattern of secured debt is limited at best. Ifredistribution is a principal determinant of secured debt, then firmswith near-perfect opportunities to engage in redistribution throughsecured debt should do so regardless of whatever other factorsmilitate against the use of secured debt. The fact that high-tort-riskfirms such as tobacco and asbestos firms have lower-than-averagesecured debt contradicts the notion that redistribution is a primemotive for secured debt. Instead, the data suggest that somethingother than redistributional motives is determining secured debt usage.B RegressionAnalysis of SecuredDebt Usage: Tobit Model

Simple statistics are instructive, but a robust identification of theeffects of redistribution on secured debt usage requires moresophisticated techniques. Section A demonstrated that firms withhigh tort risk use smaller amounts of secured debt. Although theseresults cast doubt on the redistributional theory of secured debt, theirreach is limited. To address these concerns and control for otherdeterminants of secured debt usage, I turn to regression analysis.

Previous studies of secured debt identified several non-tort-related factors affecting secured debt usage. 5 These factors includemarket-to-book ratio, earnings, marginal tax rates, size of firm, cashavailability, and country and state of incorporation. In addition tothese variables, the regressions control for the year of the observationand the exchange upon which the stock is traded. These variablescontrol for other potential differences, unrelated to tort risk, betweenhigh-tort firms and other firms.

Regressions using secured debt-to-total debt ratios or secureddebt-to-hard assets ratios have a censored dependent variable. Evenif a firm wants to hold negative amounts of secured debt, the firmwould not be able to the minimum amount of secured debt is zero.Such censoring can bias the estimates of the effects of various factors

85. See generally Barclay Smith, supra note 19 (analyzing various factors that mayimpact the type of securities firms issue, including growth and investment opportunities and taxstructures); Berger Udell, supra note 19 (analyzing the relationship between firms' credit riskand the issuance of secured debt).

2 8] 65


31/45

DUKE LAW JOURNAL [Vol. 57:1037using the Ordinary Least Squares OLS) model. To adjust for thebias, this study employs the Tobit model.&

The appropriate unit of observation for the Tobit regressions isalso a source of concern. The tort indicator variables vary acrossfirms, but they do not vary across years within the same firm; atobacco firm is always a tobacco firm. As a result, a pooled Tobitregression model treating each firm-year observation as distinct willproduce inappropriately small standard errors for the effects of tortrisk. This specification thinks there is more data than there actuallyis. To address this concern, the unit of observation for theregressions will be the mean values across years for each individualfirm. This procedure reflects the source of variance of the tort-riskvariables of interest and produces more accurate standard errorestimates. Mathematically, the specification is:

sdebti = X i/3 x tortwhere sdebt is the average (across years) secured debt-to-total debtratio for firm i X i is a vector of control variables averaged acrossyears,' tort is an indicator variable for whether or not the firm is ahigh-tort-risk firm (several different measures of this risk are used),9

86. For a discussion of the Tobit model, see ANGUS DEATON, THE ANALYSIS OFHOUSEHOL SURVEYS 85-92 1997). Note that the additional assumptions made by the Tobitmodel to correct for the bias caused by censoring, such as homoskedasticity and normality of theerror terms, are themselves suspect. See id. Indeed, sometimes the Tobit cure is worse than thecensoring disease. Furthermore, ordinary Tobit regressions treat all observations identically,ignoring potential correlations between observations of the same firm in different years. Toaddress these concerns, several procedures are employed. The standard errors reported in theregressions are estimated using a bootstrap procedure to correct for potential clustering of errorterms within firms across years. The bootstrap procedure also helps adjust standard errors forpotential heteroskedasticity. Bootstrapping, however, does not adjust point estimates forpotential heteroskedasticity biases. Professor James L. Powell's Censored Least AbsoluteDeviation (CLAD) model, by contrast, produces consistent point estimates in the face ofheteroskedasticity. See James L. Powell, Least Absolute Deviations Estimation or the CensoredRegressions Model 25 J. ECONOMETRICS 303, 303 (1984). As a result, I compare some of theTobit point estimates with estimates produced by the CLAD model. The two proceduresproduce point estimates within 10 percent of each other, suggesting that heteroskedasticity doesnot cause large biases in the Tobit estimates.

87. Because there is no variation in tort risk within firms, a fixed-effects regression model isnot feasible. For more discussion of this issue, see infraPart IV.C.1.

88. The standard error estimates for the control variables, however, will tend to beunderestimated because the regression exploits only between firm and not within-firm variation.

89. See supra text around note 85 for a discussion of the control variables.90. For a discussion of these measures, see supra Part II.

1066


32/45

2008] BANKRUPTCY 1067and i is a normally distributed, homoskedastic error term that isuncorrelated with the regressors. If the redistributional theory isempirically important (the null hypothesis for this study), then thetort coefficient 8) should be positive and significantly greater thanzero. If the redistributional theory does not have predictive power,then 8) may be zero or even negative.

Table 39 presents results of Tobit regressions using thisspecification.' Even after controlling for many other factors, Tableshows that tobacco firms use l ss secured debt than other firms. Beinga tobacco firm is associated with a desired, 93 decline in a firm'ssecured debt-to-total debt ratio of approximately .14, ceterisparibus asubstantial decrease. This estimated effect is impreciselyestimated, however. Indeed, the coefficient estimate is notsignificantly different from zero at the percent level. The results aremore significant (both statistically and economically) for the sampleof firms that ultimately declare bankruptcy as a result of asbestosliability. These firms have a desired secured debt-to-total debtratio that is .48 lower than would otherwise be expected.94

91. Results based on Tobit regressions of firm's secured debt-to-hard debt ratios(dependent variable) o the variables listed in the table. Each regression has approximately4,300 separate observations and includes other (statistically insignificant) control variables suchas the standard deviation of earnings and a dummy variable for tax carry forwards.

92. The results presented here are robust to the inclusion of many other potential controlvariables, including standard deviation of earnings, dummy variables for the presence of taxcarryforwards, and measures of abnormal earnings. Note also that the regressions do notinclude firms that are in bankruptcy because their financial data may be radically altered by thebankruptcy process. In addition, the results are robust to the use of other dependent variables.For example, using the ratio of secured debt to market value, rather than the ratio of secureddebt to total debt, does not change the results appreciably with the exception of thepharmaceutical firms' regression. This exception is not surprising given the high value ofintangibles for pharmaceutical firms.

93. Because the topic of interest is the impact of tort liability on the underlying propensityof firms to use secured debt for both firms that do and do not currently have secured debt, theordinary Tobit regressions coefficients are the appropriate coefficient of interest. Thisinterpretation applies to all of the regression results. Note that the values cannot bemechanically applied to other potential values of interest, such as the impact of a change intort's priority structure on the probability that firms will issue any secured debt. Instead, otheradjusted estimates must be used. For a helpful discussion of the interpretation of Tobitcoefficients, see generally Lee Sigelman Langche Zeng, Analyzing Censored and Sample-SelectedData with Tobit and Heckit Models 8 POL. ANALYSIS 167 1999).

94. When a dummy variable indicating whether or not a firm ever declares bankruptcyduring the time period is included in this regression, the effect of being an asbestos firm has aneven larger downward impact on secured debt amounts. This is because the average (non-asbestos) firm that declares bankruptcy has an above-average ratio of secured debt, making theasbestos companies' low secured debt usage even more exceptional.


33/45

DUKE L W JOURN L [Vol 57:1037

CA -E~.

-t 0

E

qm c

~~o00

~ jI.-

ClI.

r - COaE ~ 0C Cr .. 0

0 0 9C0u u

1068


34/45

BANKRUPTCYThis estimate is significantly different from zero at the 5 percent level,although the estimate is still quite imprecise. The imprecision of thetobacco and asbestos company tort-risk estimates is not surprising-very few companies can be clearly identified as tobacco or asbestosliability-risk companies, meaning that the small sample size limits thestatistical power of the regressions. Nevertheless, the magnitude ofthe negative effect of high tort risk contradicts the redistributionaltheory of secured debt. In addition, Table 3 presents the statistics ofthe pooled asbestos and tobacco companies to help address thelimited sample issue. When these two groups of high-tort companiesare combined, the results are both economically and statisticallysignificant. High-tort-risk companies with asbestos or tobacco liabilityhave secured debt-to-tota

Documents

Is Secured Debt Used to Redistribute Value From Tort Claimants in Bankruptcy - An Empirical Analysis